agentsclimarketplace

X post writer

Skill sergebulaev/x-skills/.codex-marketplace/x-skills/skills/x-post-writer

X (Twitter) marketing skills for Claude Code and Codex: write tweets, threads, and replies in your voice, strip AI tells, and publish via Publora. Open source, MIT.

Install
npx -y skills add sergebulaev/x-skills --skill x-post-writer

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 8 stars8 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Draft a single tweet or short auto-thread for X (Twitter) using a 2026 X hook formula (one-liner contrarian, data-point, build-in-public, mini-list, relatable cold-open), picked by goal (replies, reposts, likes, bookmarks). Respects the 280-char limit (25,000 on Premium), runs the humanizer pass, and publishes via Publora on approval. Use to write a tweet from notes. Not for long threads (use x-thread-builder) or auditing a draft (use x-humanizer --mode audit).

SKILL.md

5.2 KB, as published. Nobody here has run it

X Post Writer

Ship a single tweet, or a tight 2-3 tweet micro-thread, using hook shapes that actually travel on X in 2026. Single-tweet formulas land a complete punch in the first line, because X has no "see more" fold.

When to use

  • User says "write me a tweet about X"
  • User has a topic or a rough one-liner and wants a sharper hook
  • User wants to pick a proven single-tweet shape and fill in their voice
  • User wants a quick draft that auto-publishes on approval

Formulas this skill uses (single-tweet shapes)

CodeFormulaPrimary goalBest for
X1One-Liner Contrarianrepostsa sharp, defensible against-the-grain take
X2Data-Point Hookbookmarksone odd-precision number that reframes a thing
X3Build-in-Public Confessionrepliesa real metric from your own work, ugly ones included
X4Quote-Tweet Dunk-with-Valuerepostsadding a layer to someone else's take (quote tweet)
X5Mini-List Tweetbookmarksa scannable list that fits in one tweet
X6Relatable Cold-Openlikesa specific shared moment, no setup
X11Third-Person Case-Study Flexrepostssomeone else's absurd, specific result as anyone-could-copy bait

Full skeletons in ../../references/hook-formulas.md. For long-form tweetstorms (X7-X10), use x-thread-builder. X11 was validated as the top-recurring winning shape in a ~450-tweet corpus (July 2026); prefer it over a first-person "How I" when the result you are citing is someone else's.

Pick by goal first

GoalReach for
RepliesX3, X1
RepostsX1, X4, X11
LikesX6
BookmarksX2, X5, X11

Steps

  1. Gather inputs. Topic, angle, any rough draft, target audience (builders / founders / a niche), and the goal (replies / reposts / likes / bookmarks).
  2. Pick the container. If the idea is one claim, one number, or one moment, it is a single tweet. If it is a scannable one-line-per-item list, still a single tweet (X5). If items need teaching or the idea has a build, hand off to x-thread-builder.
  3. Pick the formula. Use the goal table to shortlist, then suggest 2-3 that also fit the topic and let the user choose.
  4. Draft the tweet. Fill the skeleton in the user's voice. Respect the 2026 X rules:
    • First line carries the whole load (no fold).
    • Under 280 chars on a standard account (emoji count as 2 chars). Confirm the user's tier; only go long if they are on Premium and the length earns it.
    • One idea per tweet. Line breaks as beats.
    • 0-1 hashtag at the end, 0-1 emoji, none on a serious take.
    • No external link in the tweet (offer to put it in a reply).
  5. Humanizer pass. Strip em dashes, AI vocab, rule-of-three, generic openers. Add a specific number or named entity where the claim allows it.
  6. Optional audit. Invoke x-humanizer --mode audit for a pass-fail check.
  7. Approval card. Show: formula used, full draft, char count (note if emoji push it over 280), suggested posting window, primary goal.
  8. On approval. Call lib.publish(kind="post", draft_text=<approved>, target_url="https://x.com/compose/tweet", platforms=[<X_PLATFORM_ID>], scheduled_time=<iso_or_None>). The wrapper handles Publora / manual / diy routing. If the content runs long, Publora auto-splits it into a numbered thread.

Hard rules

Global voice rules: see root SKILL.md Voice rules. Additional skill-specific rules:

  • The first line must stop the scroll on its own. Rewrite any opener that needs line 2 to make sense.
  • Keep a standard-account tweet inside 280 chars. If it does not fit, tighten it or escalate to a thread, never ship a truncated thought.
  • One specific number where the claim allows it. 2.4x beats "way better".
  • Do not hard-sell the user's product. One natural mention max.

Anti-patterns (skill will refuse)

  • ALL CAPS first line for intensity. Carry intensity with word choice.
  • Em dashes anywhere.
  • "Unpopular opinion:" on a take that is actually popular.
  • Padding a one-line idea with filler to look substantial.
  • Rule-of-three lists without specifics.
  • "leverage", "fundamentally", "game-changer", "deep dive".
  • External link in the tweet body.
  • Engagement bait ("RT if you agree", "reply YES").

Resources

  • ../../references/hook-formulas.md - all 11 X formula skeletons (X1-X6 and X11 are single-tweet)
  • ../../references/algorithm-heuristics.md - 2026 X ranking rules (signals, timing, limits)
  • references/single-tweet-checklist.md - the per-tweet scrub and fit list

Related skills

  • x-thread-builder - when the idea needs more than a tweet
  • x-humanizer - aggressive AI-tell scrubber, plus --mode audit for review
  • x-hook-extractor - reverse-engineer a hook from a tweet you admire

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.